[PATCH v3] firmware: coreboot: Don't register a pdev if screen_info data is present

[Javier Martinez Canillas]

Thomas Zimmermann <tzimmermann at suse.de> writes:

Hello Thomas,

> Hi Javier
>
> Am 16.09.24 um 10:36 schrieb Javier Martinez Canillas:
>> Thomas Zimmermann <tzimmermann at suse.de> writes:
>>
>> Hello Thomas and Tzung-Bi,
>>
>>> Hi
>>>
>>> Am 15.09.24 um 09:44 schrieb kernel test robot:
>>>> Hi Javier,
>>>>
>>>> kernel test robot noticed the following build errors:
>>>>
>>>> [auto build test ERROR on chrome-platform/for-next]
>>>> [also build test ERROR on chrome-platform/for-firmware-next linus/master v6.11-rc7 next-20240913]
>>>> [If your patch is applied to the wrong git tree, kindly drop us a note.
>>>> And when submitting patch, we suggest to use '--base' as documented in
>>>> https://git-scm.com/docs/git-format-patch#_base_tree_information]
>>>>
>>>> url:    https://github.com/intel-lab-lkp/linux/commits/Javier-Martinez-Canillas/firmware-coreboot-Don-t-register-a-pdev-if-screen_info-data-is-present/20240914-053323
>>>> base:   https://git.kernel.org/pub/scm/linux/kernel/git/chrome-platform/linux.git for-next
>>>> patch link:    https://lore.kernel.org/r/20240913213246.1549213-1-javierm%40redhat.com
>>>> patch subject: [PATCH v3] firmware: coreboot: Don't register a pdev if screen_info data is present
>>>> config: riscv-randconfig-001-20240915 (https://download.01.org/0day-ci/archive/20240915/202409151528.CIWZRPBq-lkp@intel.com/config)
>>>> compiler: clang version 15.0.7 (https://github.com/llvm/llvm-project 8dfdcc7b7bf66834a761bd8de445840ef68e4d1a)
>>>> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240915/202409151528.CIWZRPBq-lkp@intel.com/reproduce)
>>>>
>>>> If you fix the issue in a separate patch/commit (i.e. not just a new version of
>>>> the same patch/commit), kindly add following tags
>>>> | Reported-by: kernel test robot <lkp at intel.com>
>>>> | Closes: https://lore.kernel.org/oe-kbuild-all/202409151528.CIWZRPBq-lkp@intel.com/
>>>>
>>>> All errors (new ones prefixed by >>):
>>>>
>>>>>> ld.lld: error: undefined symbol: screen_info
>>>>      >>> referenced by framebuffer-coreboot.c:27 (drivers/firmware/google/framebuffer-coreboot.c:27)
>>>>      >>>               drivers/firmware/google/framebuffer-coreboot.o:(framebuffer_probe) in archive vmlinux.a
>>>>      >>> referenced by framebuffer-coreboot.c:27 (drivers/firmware/google/framebuffer-coreboot.c:27)
>>>>      >>>               drivers/firmware/google/framebuffer-coreboot.o:(framebuffer_probe) in archive vmlinux.a
>>> Not all platforms define screen_info. Maybe fix this by following
>> Yes, after reading the build errors reported by the robot I remembered
>> that we had similar issues with sysfb, for example commit 1260b9a7020
>> ("drivers/firmware: fix SYSFB depends to prevent build failures") fixed
>> one of those.
>>
>>> Tzung-Bi's advice of removing the local variables and then guard the
>>> test by CONFIG_SYSFB. If SYSFB has been defined, screen_info has to be
>>> there. It's not a super pretty solution, though.
>>>
>> If possible I would prefer to avoid the ifdefery in the driver. I also
>> believe that the local variables makes the code easier to read. But if
>> you folks think that's better to drop them, I can do it in the next rev.
>>
>> Another option is to restrict the architectures where this driver could
>> be build. As far as I understand it is mainly for x86 and ARM64 arches.
>>
>> These two have a screen_info defined so the driver will build correctly.
>> I can include a preparatory patch that adds a "depends on x86 || ARM64".
>
> That feels arbitrary, as the dependency is not really in coreboot, but 
> in sysbf. What you'd want is a HAVE_SCREEN_INFO define on the 
> architectures that provide it. IIRC earlier attempts to add this have 
> failed. :/
>
> If you don't want the ifdef-ery in coreboot.c, you could add a helper to 
> sysfb. Let's say
>
>    bool sysfb_handles_screen_info(void)
>
> returns the result of the test. If you put it next to sysfb_disable(), 
> you could add an empty wrapper into the sysfb.h header as well. [1]
>
> (There's still the possibility that screen_info is available, but sysfb 
> has been disabled. But that's not different from how it currently works.)
>

I like that. And when SYSFB is not enabled, sysfb_handles_screen_info()
could just be defined as "return false;" which will indicate to drivers
that would have to handle the system frambuffer themselves.

I'll try to type the patches later today.

-- 
Best regards,

Javier Martinez Canillas
Core Platforms
Red Hat